CVMar 20, 2024

VL-Mamba: Exploring State Space Models for Multimodal Learning

arXiv:2403.13600v1121 citationsh-index: 16ENLSP
Originality Incremental advance
AI Analysis

This work addresses computational efficiency for researchers and practitioners in multimodal AI, though it is incremental as it adapts existing state space models to a new domain.

The paper tackles the high computational cost of Transformer-based multimodal large language models by proposing VL-Mamba, which uses state space models for multimodal learning, achieving competitive performance on diverse benchmarks.

Multimodal large language models (MLLMs) have attracted widespread interest and have rich applications. However, the inherent attention mechanism in its Transformer structure requires quadratic complexity and results in expensive computational overhead. Therefore, in this work, we propose VL-Mamba, a multimodal large language model based on state space models, which have been shown to have great potential for long-sequence modeling with fast inference and linear scaling in sequence length. Specifically, we first replace the transformer-based backbone language model such as LLama or Vicuna with the pre-trained Mamba language model. Then, we empirically explore how to effectively apply the 2D vision selective scan mechanism for multimodal learning and the combinations of different vision encoders and variants of pretrained Mamba language models. The extensive experiments on diverse multimodal benchmarks with competitive performance show the effectiveness of our proposed VL-Mamba and demonstrate the great potential of applying state space models for multimodal learning tasks.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes